Which: High or Low Flowrate?

[ymboc]

Hey guys... long time reader, first time poster :P

anyway.... I've read a lot of posts, articles and the like. Some say you need high flow rate, others say low flow rate is better, others say that high flow rate is nice but isnt any better than low flow rate.

The trend I've seen lately (is it my imagination?) is to say that low flow rate is better but then at the same time I see that when someone asks 'which pump?' the common answer is the eheim 1250 -- which isnt exactly low flow rate.

So as it pertains to a 'generic' setup (HeaterCore + Maze3/Spiral Design WB). Which is better?

Thanks
-ymboc

[Cyco-Dude]

it depends on the components you use. from what ive read so far, a system like that would benefit from higher flow as opposed to lower flow.

[webmedic]

Well this is tricky one. I've used pumps in my test setups from about 60gph to about 450gph both inline and submerged and if you are using a heater core low flow rate is fine. The only reason for high flow rate is if you are using a mulitpass high drop radiater like the old dd super cube which I also have. that style of radiator requires high flow because it will kill your flow.

As to the question on pumps that is also wrong an ehiem is good ( I havethe 1250 ) but It's not always needed. The only pumps to always stay away from are the dc pumps. They will die whithin a few months if used 24/7.

[SCompRacer]

It depends on components and what results you will accept. It seems that the Maze2 and Maze3 like high flow even with a less restrictive heater core or the BIX. Actual flow from an Eheim 1250 with a Maze2, 6 x 6 heater core and 1/2 lines is 95GPH. I dropped the line size to 3/8," actual flow drops to 64GPH and my CPU temp goes up 1C. Using restrictors to slow the flow to 55GPH gets me another 1C increase in temps.


While I haven't flow tested it yet, replacing that 3/8" ID outlet on the 1250 with a modded fitting that gets me to 1/2" without a 3/8" adapter resulted in a 1C decrease in loaded CPU temps.

So if you want the best temps possible, increased flow with certain components is the way to go. If you just want acceptable temps, less flow is fine.

[webmedic]

Thats not true either I've tested at least 6 pumps on my maze 2 and the flow doesn't make that big of difference. The bigger pumps also produce more heat. It's dimminishing returns. I have a spreadsheat that I started and fans and rads are what make the most difference not pumps and the water blocks don't care as much about the flow as the rads do.

[SCompRacer]

Well I've done some testing too and it is very true. First off, if you run a larger pump and restrict the flow, the pump will run warmer and can transmit more heat into the outlet flow than if it runs at unrestricted flow.

If you don't have a small fan to cool the pump, it can run 28C or higher in 21C ambient. If you use a 22CFM fan to cool the pump, you can reduce the pump temp to 25.6C. My outlet versus inlet temp is only .4C higher like this.

If I restrict any flow from the pump, my Dig Doc5 sensor touching the CPU core goes up within seconds, release the restriction and it goes back down.

As I stated, reducing flow causes temps to go up 1 to 2C with the same system, same overclock, same load, same pump in the same controlled ambient conditions, so it is very true.

[pHaestus]

The idea that a waterblock would not benefit from increasingly higher flow is not correct: forced convective heat transfer is directly proportional to the velocity of air or water.

The idea that a particular water cooling system may not benefit from higher flow rates is correct: there is a balance between pump heat, radiator performance, and waterblock performance to be optimized. Here is more info:

http://www.overclockers.com/articles481/

[Kevin]

Originally posted by SCompRacer
So if you want the best temps possible, increased flow with certain components is the way to go.

Just as electrical current is constant in a loop, so is flow rate of water in a loop.
-kevin

[Brad]

kev, I think he was meaning that a maze3 + heatercore will like more flowrate than a gemini + BI.


I prefer high flow rate, simply you can always add more components and not feel restricted on what you can do before you run out of pump power. Or run more 90 degree bends, or whatever

[ymboc]

Very nice replies guys! Actually, I've read the article on heatercore vs flowrate performance over at overclockers.com a number of times before posting, but was still somewhat baffled over the contraditions I've seen in the forums (obviously not just here). Thanks for clearing that up.

Cheers!
-ymboc

[webmedic]

My above comment was only in regards to the maze2.

Sorry to qualify my post it's not that flow rate does not matter at all. The problem is just as stated above any restriction in the system creates more heat. Also using verry large pumps produce more heat than small ones. There is a point of diminishing returns. Not that larger pumps aren't better in some ways but they also make more heat. There is also a point when a pump is to small. Personnaly I've found pumps in the 130-150 gph to be the best mix of performance verses heat. All the rest is bad design. If you set up your system for max flow the the pump is less of an issue.

That is why I use 1/2 tubing throughout my system and 5/8 fitting. The 1/2 fittings are 1/4 on the inside or 3/8 at best. This restricts flow. Using a heater core over a dd super cube maximizes flow. Having a high flow block with lots of surface area increases flow. I can give many examples.

The important thing is to have max flow in your entire system( good system design ). Not a system that restricts flow ( bad system design ). With a high flow system less heat is produced by the pump and a smaller pump can be used. But even still some of the worst setups I had usually gave temps whithin 2c with different pumps.

Here is were the diminishing returns come in. larger pump = more wattage. Take a look at the ehiem 1250 it's about 28 watts even if all that is not transfered directly to the water it's still allot more than my 130gph pump which is only 6 watts. Also the ehiem has a small hole on the inside that vents water to the inside of the pump around the turning shaft this is how the ehiem is cooled. So the point of submerged or not is also not an issue.

[pHaestus]

Originally posted by Kevin

Just as electrical current is constant in a loop, so is flow rate of water in a loop.
-kevin

But not water velocity. Increased water velocity improves waterblock performance; it is just that the simplest "blanket" approach to higher water velocities is higher flow rates.

[Brians256]

If I might amplify on what pHaestus said, velocity is definitly not constant throughout a whole watercooling system.

If you narrow the channel, velocity increases to maintain the same flow rate. If you widen the cross-section, the flow rate drops. Water does not compress (not enough to be noticable at the macro scale). This is why water is slower through a heater core: it has a huge cross section compared to the tubing.

On the other hand, trying to use a narrow nozzle in the waterblock to increase velocity will only work if you have a very powerful pump. Mag-drive pumps cannot produce the pressure necessary to effectively drive something like that. The pumps that can produce that much pressure are, if I guess correctly, much noisier.

It's all tradeoffs.

[Haddy]

it really all comes down to what u got....nobody can say for sure (unless they have tested it out the azz and u get the same setup) what will work best......but pretty much my guess is that most blocks like half way decent flow....but then u get blocks like the low flow gemini that like lowww flow....so really its all about what u buy

[Brians256]

Haddy, do you have an URL for someone that tested the Gemini block and found that it did better with low flow? I'd be VERY surprised if it did. On the other hand, as part of a complete system, I can see the system doing better at lower flow rates (if the radiator does poorly with higher flow rates).

I just can't imagine a waterblock doing better with lower flow rates, if the inlet water temperature is constant for all tested flow rates. That is a very big "if", though.

[pHaestus]

Haddy: reread what I said. No waterblock "likes" low flow over high flow. I challenge you to find one set of reliable data that proves that waterblock performance improve with lower flow rates. Any observed decreases in performance with increasing flow rates are due to the radiator; not the block.

[Kevin]

if narrow channels = high velocity, then high velocity = less surface area so i really think everything depends on the person's individual setup. Haven't we had a thread like this before?
-Kev

[pHaestus]

A person's individual setup can't change the fundamentals of heat transfer. Look at this:

http://www.efunda.com/formulae/heat_...iew_forced.cfm

Note the delta T term (block surface temperature vs. water temperature in our case). There is no way to increase this term (and enhance heat transfer rates) other than to increase flow through the block. What about that h term though (the convection heat-transfer coefficient)? This is related (among other things) to the thickness of the boundary layer of water on the surface of the block. Here is a nice overview:

http://chemeng1.kat.lth.se/staff/ulf_b/b3_heat.htm

Increased water velocity (and therefore turbulence) will decrease (but not eliminate) the boundary layer of water at the surface. This improves heat transfer markedly compared to laminar flow through a larger boundary.

For a given block, performance will always increase with flow rates. However, in a complete cooling system the radiator and pump will not and so there exists a "sweet spot" where flow rate is optimized for the components that are used. I said this in the beginning btw.

[Kevin]

pH - I think you misunderstood what I said. I agree with you about the velocity thing. But velocity in one point of any loop is constant. On other words, velocity through a 1 inch segment of a hose will be constant (minus bends, etc). To increase the velocity, you can pinch the line. Velocity will increase through that area while overall flow will decrease. BUT you can't pinch a waterblock. Thus, making smaller channels would increase the velocity, but lower the surface area. That's all I was saying. That there is a sweet spot there as well. And yes, I agree with you in that an increase in flow rate will never hurt waterblock performance, and usually help (until the point of diminishing returns). The converse of that statement holds true with radiators so in that sense, there is a sweet spot. um... basically what you said
-Kev

[Haddy]

Originally posted by Brians256
Haddy, do you have an URL for someone that tested the Gemini block and found that it did better with low flow? I'd be VERY surprised if it did. On the other hand, as part of a complete system, I can see the system doing better at lower flow rates (if the radiator does poorly with higher flow rates).

I just can't imagine a waterblock doing better with lower flow rates, if the inlet water temperature is constant for all tested flow rates. That is a very big "if", though.

http://www.overclockers.com/articles546/ i consider that a low flow system considering the pump....thats not my fav wb review but its still the only one ive seen with a "low" power pump

[Haddy]

Originally posted by pHaestus
Haddy: reread what I said. No waterblock "likes" low flow over high flow. I challenge you to find one set of reliable data that proves that waterblock performance improve with lower flow rates. Any observed decreases in performance with increasing flow rates are due to the radiator; not the block.

its not rated for anything above 200gph, so if ur waterblock's seals break then i know it wont "like" that....least thats what i get outa that block being rated for <200gph

[pHaestus]

Haddy the fact that one block outperforms another at a particulr flow rate doesn't at all prove that blocks perform better at lower flows.

As for the 200 gph being the point that o rings break; I wouldn't purchase anything from a manufacturer of such a product. It must be held onto the top by positive thoughts and butterfly wings for it to be that sensitive...

[Brians256]

I'm very happy that the Gemini performs well at that low of a flow rate, but I really believe that it would perform even better at a higher flow rate. As pHaestus points out, increased flow rate will help a waterblock perform in several ways.

1. Increased velocity causes the boundary layer to thin out, which means that the heat has less distance to travel before it can get "carried away" by the moving water.

2. Increased velocity means that the water in the block is more constantly refreshed with the cooler incoming water.

3. Increased velocity causes the water within the block to "mix up" more (similar but not the same as the boundary layer). This is "turbulent flow" versus "laminar flow".

On the other hand, if the water moves too fast, it could hit the copper so hard that it etches it. Somehow, I doubt that our puny mag-drive pumps are that powerful.

[redleader]

So as it pertains to a 'generic' setup (HeaterCore + Maze3/Spiral Design WB). Which is better?

Higher flow, but the difference often isn't huge between given flow rates (1250 with 1/2 vs 1250 with 3/8 for example).

[Brad]

just go high flow rate with almost all setups